Algorithm of Friend Recommendation in Online Social Networks Based on Local Random Walk

Online social networks(OSNs) have become popular,which provide users with a new communication and information sharing Internet platform.In OSNs,Recommending friends to registered users is a crucial task.On the one hand,OSNs often recommend friends for users based on local-based features of the social graph(i.e.based on the number of common friends that tho users share).This method considers only pathways of lenght 2 between users and does not exploit all different length paths of the network and other information.On the other hand,there are global-based approaches of friend recommendation in OSNs which detect all pathway structures of the network.But its computation cost is quite high for large scale OSNs.In this paper,we propose a new approach of friend recommendation in OSNs which traverses all the paths of limited length through randomwalk based on "small world" hypothesis.This new method provides users with both fast and accurate friend recommendation in OSNs.To demonstrate practical applicability of the new aproach,we use two real world datasets to evaluate our novel approach.Experimental results showed the approach can significantly improve the accruracy of friend recommendation in OSNs.